Performance assessment and design optimisation of VRPM (transverse flux) machines by finite element computation

The work is concerned with studies of a novel form of electric machine, the variable reluctance permanent magnet (VRPM) or transverse flux machine. The main thrust of the present work is to establish the limits of performance of VRPM machines, with particular reference to specific torque, for the principal alternative machine geometries that are found to be of interest. This is done by use of advanced finite element computational applications* both two-dimensional and three-dimensional, together with the development of analytical approaches to particular matters.One machine configuration, selected for intensive study, is a prototype built at the University of Southampton, which is topologically a single-sided radial-gap surface-magnet machine with inverted structure. The prototype as built achieves 27.5 kNm/m3, which is about 4.9 times greater than for an industrial induction motor. Extensive three-dimensional design optimisation studies have established that this performance can be improved by about 15% by appropriate changes to machine proportions.The work has shown for this type of machine, in a form applicable to machines of differing size, the optimum proportions of tooth-width, tooth-pitch and gap. Two-dimensional field computation has also been used, to produce simplified look-up tables that can be used in design optimisation routines. This work is found to lead to results closely agreeing with the three-dimensional studies, although absolute values of torque and other parameters, calculated by two-dimensional methods, are considerably in error, because the VRPM machine is strongly three-dimensional. This matter is explained.Three-dimensional optimisation studies have also been made of a double-sided, axial-gap, flux-concentrated machine configuration, related to a prototype machine built at the University of Newcastle Upon Tyne. These studies show that the specific torque of the Newcastle machine can be increased by at least 30% by optimising the detailed proportions. They also confirm that this configuration is inherently the better of the two in terms of specific torque (although it is a more complicated and less robust structure), by a factor which can be as high as 1.64, depending upon what is taken as a fair basis of comparison. The underlying reasons for this are explained.VRPM machines suffer from low power factor, and the underlying reasons for this have been investigated. It is shown a single figure of merit accounts for this feature, and also makes easy comparison possible between VRPM and conventional synchronous machines. It becomes clear that the low power factor is an unavoidable feature, and that it is contributed to by the remarkably low flux utilisation that is shown to be a feature of VRPM machines, both in regard to stator-excited and magnet-excited flux components